An image-taking apparatus which reduces occurrence of unnatural focus changes is disclosed. The image-taking apparatus includes a first detector which outputs a first signal representing a focus state of an image-taking optical system based on a predetermined-frequency component of an output signal of the image-pickup device and a second detector which outputs a second signal different from the first signal. The image-taking apparatus also includes has a controller which performs a first focus control for controlling drive of the focus lens based on the first signal and a second focus control for controlling drive of the focus lens based on the second signal. The controller performs the second focus control in a case where an in-focus state cannot be achieved with the first focus control.